Do a paid internship at Docear in Germany, summer 2014 (US, UK, and CA Bachelor students only)
Like the previous years, we offer the opportunity for Bachelor students from the US, UK or Canada to do a paid internship at Docear in summer 2014 (if you are from Germany, please read here). The internship should last for 8-12 weeks with the earliest start date being May and the latest being August. You will be paid 650 Euros a month, a 160 Euro allowance for travelling, and health insurance. International travel costs will not be covered. You will be placed with Docear’s core team in Magdeburg, close to Berlin. So, if you love Docear as much as we do, are a passionate software developer or statistician (or want to become one), and are interested to spend your next summer in sunny Germany, read on.
Your project will be to improve Docear’s recommender system. As such, it will be your task to support the Docear team in researching how the interests of Docear’s users can be identified from the data, and how these interests can be matched with research papers. You will do literature research, create new ideas, analyze the users’ data, and implement new recommendation approaches in JAVA. Of course, you don’t have to do all of this alone – you will be closely cooperating with us, the Docear team. Your work will be integrated into Docear and used by thousands of researchers around the world. If you are interested you are also very welcome to write a research paper with us, or, if you home university allows this, use your internship to work on your Bachelor’s thesis. If you have other ideas than improving the recommender system, please let us know. We are open to new ideas and you may do anything you want as long as it will make our users happy.
If you are primarily interested in software developing you should have a profound knowledge of the programming language JAVA. Knowledge in statistics, machine learning, other programming languages (especially C/++ or Python) and/or MySQL, neo4j, Hibernate, Jersey, REST Web Services, Tomcat, and Apache is a plus, but not required. If you are primarily interested in research, i.e. analyzing the data we have, you should have profound knowledge in statistics and in at least one statistic software tool (e.g. SPSS, R, …).
The internship is offered in cooperation with the German Academic Exchange Service (DAAD). Therefore, if you are interested in the scholarship, apply before January 31, 2014. To apply, read the application guidelines, register, log-in, browse DAAD’s internship database, find Docear’s project page, and apply.
Even if you are not eligible to apply, or just not interested, please tell your friends of the internship opportunity. This is really important. Only when a significant number of students applies, DAAD will sponsor the project. Therefore, please promote the internship opportunity as much as you can.
Here are some of the ideas
Academic Recommender System
Brief Description: One of our main goals is recommending relevant papers, journals, conferences, universities, etc. to our users. To give relevant recommendations we need detailed user models and there are various approaches to do so (content based and collaborative filtering). However, classic user modeling approaches usually focus on modeling interests based on websites, emails, etc. but not on mind maps as Docear is using them. Therefore, it will be your task to find out which existing user modeling approaches are most suitable for mind maps and how they need to be adjusted to be most effective.
This project is a major project which will require lots of time. If you want to do this project as part of an internship of Bachelor/Master thesis, we may break it down to smaller parts. For instance,
You could focus on applying standard user modeling approaches to mind maps. These would be used as a base-line to evaluate the effectiveness of the following approaches
You could focus on the suitability of the terms in a mind map for building a user model and matching them with items (e.g. academic articles)
You could focus on the links and citations in a mind map for building a user model and matching them with items (e.g. academic articles)
You could focus on collaborative filtering, i.e. determining similar users, and recommending items of these similar users
You could focus on combining the previous approaches (of course only after they are implemented)
Expected Results: A Java library that creates user models based on mind maps and matches the user models with appropriate items to recommend.
Expected Research Results (optional): Evaluation of how effective your user modeling library works, based on click through rates or a user study.
Required Knowledge: Java, Databases (MySQL and ideally Neo4j), Text Mining and Information Retrieval (recommended), Machine Learning (recommended), User Modeling (recommended)
Mind Map Analysis (Pure research project)
Description: Mind maps apparently differ in their structure and way of creation from emails, academic papers, or web pages. But what exactly are the differences? Your task will be to analyse the thousands of mind maps we collected from our users and to find out what makes them unique. To do so, you will, among others, compare mind maps with web pages and academic articles (it will be up to you to which other documents you compare mind maps with and where to get these other documents from). Your work will be the ground work for our academic recommender system (see above) because to provide good mind-map based recommendations you need to know the unique features of mind maps.
Expected Research Results: A study about the unique features of mind maps, compared to web pages, academic articles, and maybe emails, social tags and search queries.
Required Knowledge: A programming language to analyse the data (preferably Java, C/++, or Python), XML, some MySQL and ideally Neo4j
Description: Docear is a great software with lots of unique features but to be honest, it’s not very user friendly. Your task is to improve the usability experience of Docear. Find out what disturbs the users and how to make features more accessable and easier to use.
Expected Results: A new version of Docear that is easier to use than the current version
Expected Research Results (optional): A study to find out how much better your new version exactly is. You could either perform a user study (e.g. give certain tasks to users and analyse how long they need for it) or analyse the usage behavior (we monitor which functions users use how often and how long) and find out whether users of your new version use more feature (or need less time to do certain tasks).
Required Knowledge: Excellent Java skills, knowledge in user interface design, OSGi (recommended) and it would certainly help if you had used Docear for a while to know the workflow.